BULLETTIME, the viewer turn around the still subject, the character is frozen in time.
Description
FAQ
Comments (17)
Luisa, come on, you need an H100 urgently.
why
Hahahaha
@LDWorksDavid I DONT GET IT 😭
@premiumfp2025958 yeah, currently i'm wasting my own money just for fun to make theses loras in runpod, i love doing this.
@luisa_pinguin I think is because you were saying you were training with 1 image/1 video for fastest training possible and low VRAM usage too so maybe he is trying to say that you need a more powerful GPU to train "normally". That's at least how I understood it.
@LDWorksDavid no, that was trained on at least 10 videos i think
@luisa_pinguin Oh! then my bad. I was still curious about your 1 video method, you should do some comparisons on training maybe 4 vs 1, etc. Theorically will always be more the better (till some point).
how to work
TY @luisa_pinguin , Please make one about a food competition, bizarre Olympics, or kung fu fights.
Thank you for your dedication, your designs are very cool, keep it up!
you save my life
Amazing, no need for complex description, and there are only files this small. Only about one-tenth of other LoRA.
Thanks for sharing! is there a place to download the training data? Really curious about the data set and captioning.
Can it be that this model doesn't work very well with lightx2v? I really like the concept, thanks for that! The output for me is pretty low quality.
Luisa, AMAZING work! Clean,efficient and gorgeous! Thank you.
Curious,is there a way to control the direction Counter Clockwise or Clockwise?
Thank you!
Details
Available On (1 platform)
Same model published on other platforms. May have additional downloads or version variants.